Neoadjuvant therapy in primary liver cancer: From the missing piece to a new paradigm

Neoadjuvant therapy has long been the missing piece in the treatment of primary liver malignancies. Despite the well-recognized burden of postoperative recurrence, the field has lacked robust randomized evidence to support perioperative intervention, and most available data have been retrospective, single-arm, or early-phase.1 As a result, upfront surgery has remained the default curative pathway, while the potential roles of preoperative systemic therapy in eradicating occult disease, testing tumor biology, and refining surgical selection have remained largely unproven. Recently, 2 clinical trials have separately demonstrated the efficacy of perioperative or neoadjuvant therapy in hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (iCCA), the 2 major forms of primary liver cancer. CARES-009 showed that perioperative camrelizumab plus rivoceranib significantly improved event-free survival in patients with resectable HCC at intermediate or high risk of recurrence.2 ZSAB-neoGOLP showed that neoadjuvant GOLP—gemcitabine, oxaliplatin, lenvatinib, and toripalimab—significantly prolonged event-free survival in patients with high-risk resectable iCCA.3 Together, these trials represent a landmark step toward evidence-based neoadjuvant therapy in primary liver cancer. CARES-009 may represent a turning point for perioperative therapy in resectable HCC, but it also raises an unresolved question central to the field. This randomized phase 2/3 trial was designed to test a perioperative strategy as a whole, rather than to dissect the relative contributions of its neoadjuvant and adjuvant components. This distinction matters. In HCC, the adjuvant setting has so far yielded mixed or disappointing results, with the initial recurrence-free survival benefit seen in IMbrave050 not maintained on updated analysis.4 Preclinical evidence from a recent mouse study suggests that surgical resection itself may transiently impair antitumor immunity, potentially creating a postoperative window that limits the efficacy of adjuvant anti–PD-1 therapy.5 Thus, CARES-009 should not be interpreted simply as proof that the full perioperative continuum is required, but rather as the first high-level randomized evidence that systemic therapy given before surgery can alter outcomes in resectable HCC. The next challenge will be to determine whether postoperative continuation is necessary for all patients, or whether the major benefit of this strategy is already concentrated in the preoperative phase. ZSAB-neoGOLP also raises a question that extends beyond efficacy: how much treatment burden is justified in the neoadjuvant setting? The GOLP regimen was developed from advanced and conversion settings, where maximizing tumor response is a prerequisite for achieving resectability.6 In that context, such broad combination therapy is biologically and clinically intuitive. However, in high-risk yet resectable iCCA, the goal of neoadjuvant treatment is different. In ZSAB-neoGOLP, neoadjuvant GOLP significantly prolonged event-free survival, but R0 resection rates were already high and largely similar between groups. This suggests that the benefit of GOLP may lie less in rendering surgery feasible than in providing early systemic control and biologic selection. The question, then, is not whether GOLP can work, but whether such an intensive 4-drug regimen is necessary for all patients with resectable high-risk disease. The future challenge may not simply be whether to intensify neoadjuvant therapy, but whether such intensification can be deployed selectively through biomarker-guided patient stratification, including pre-treatment liquid biopsy approaches to identify patients most likely to achieve meaningful responses.7 What should follow CARES-009 and ZSAB-neoGOLP is not simply a longer follow-up, but a deeper biological resolution. Across HCC and iCCA, the next major advance will depend on how precisely clinical trials can interrogate patient-derived samples and treatment-induced biological states, rather than relying on efficacy endpoints alone. Such information is difficult to fully recapitulate in animal models, particularly when the relevant questions involve interpatient heterogeneity, dynamic immune remodeling, and paired pre-treatment and post-treatment changes under therapy. In this regard, the previous multimodal analysis of GOLP-treated iCCA is especially instructive: by integrating paired tumor specimens with single-cell transcriptomics, TCR sequencing, multiplex immunohistochemistry, and functional validation, the investigators showed that response and resistance were associated with distinct immune ecosystem states, providing a biological framework that could not have been inferred from survival curves alone.8 Future perioperative trials in primary liver cancers should therefore be designed not only to test whether a regimen works, but also to define why it works, for whom, and at what treatment burden, through embedded paired tissue sampling, pathological response assessment, circulating biomarkers, and spatial analyses, including spatial omics approaches to resolve tissue architecture and pathology-based quantification of immune cell composition.9 Another important frontier is whether neoadjuvant therapy can serve as a bridge to liver transplantation in selected liver malignancies, particularly cholangiocarcinoma, where transplant pathways are being re-examined in the era of modern systemic therapy.10 Indeed, ClinicalTrials.gov indicates that the field is rapidly gaining momentum, with nearly 100 prospective neoadjuvant or perioperative studies in primary liver cancer currently recruiting or preparing to open. These trials span a broad range of clinical scenarios, from neoadjuvant treatment before local therapy in early-stage disease to intensified perioperative approaches for tumors at high risk of recurrence and bridge-to-transplant strategies (Table 1). Importantly, this momentum should not be viewed simply as an extension of conventional downstaging strategies. Whereas downstaging is primarily intended to convert tumors initially outside curative criteria into candidates for resection, transplantation, or other local therapies, neoadjuvant therapy has the broader goal of treating occult systemic disease, probing tumor biology in vivo, and refining selection before a planned curative-intent intervention. The next frontier may therefore be not only intensified perioperative treatment for biologically high-risk tumors, but also the prospective expansion of true neoadjuvant strategies to selected patients with intermediate-stage disease, and perhaps even to earlier-stage disease before local therapy. In that setting, future neoadjuvant studies in liver cancer should serve not merely to test efficacy, but as platforms for biologic annotation, mechanistic discovery, and clinically informed refinement. TABLE 1 - Selected registered neoadjuvant and perioperative clinical trials in primary liver cancer Trial ID Operation Regimen Phase Hepatocellular carcinoma NCT04727307 Ablation Atezolizumab + Bevacizumab Phase II NCT04954339 Resection Atezolizumab + Bevacizumab Phase II NCT05027425 Transplantation Durvalumab + Tremelimumab Phase II NCT05185505 Transplantation Atezolizumab + Bevacizumab Phase IV NCT05185531 Resection SBRT + Tislelizumab Phase I NCT05920863 Resection Lenvatinib + Tislelizumab + TACE Phase II NCT06812039 Resection HAIC + Donafenib + Sintilimab Phase II NCT06884982 Resection QL1706 + Lenvatinib Phase II NCT06045975 Ablation Durvalumab + Tremelimumab Phase II NCT07475026 Resection Tislelizumab + Lenvatinib Phase III NCT06664996 Resection Sintilimab + SBRT Phase II NCT07018947 Resection Atezolizumab + Bevacizumab Phase II NCT07268131 Resection QL1706 + TACE Not applicable NCT04658147 Resection Nivolumab + Relatlimab Phase I NCT07154082 Resection Apatinib mesylate + Camrelizumab + TACE Phase II NCT06904014 Resection Sintilimab + Lenvatinib + HAIC Phase II NCT05701488 Resection Y90 SIRT+ Tremelimumab + Durvalumab Phase I NCT04777942 Not applicable TACE + HAIC Not applicable Intrahepatic cholangiocarcinoma NCT05672537 Resection Durvalumab + Gemcitabine + Cisplatin Phase II NCT06050252 Resection Durvalumab + Gemcitabine + Cisplatin Phase II NCT05967182 Resection Pembrolizumab + Gemcitabine + Cisplatin Phase II NCT06208462 Resection HAIC + Lenvatinib + Adebrelimab Phase II NCT06296563 Resection Adebrelimab + Apatinib + HAIC Phase II NCT06721286 Resection Toripalimab + Gemcitabine + cisplatin Phase II NCT06903273 Resection Tislelizumab + Gemcitabine + Cisplatin + S-1 Phase II NCT06862934 Transplantation Durvalumab + Gemcitabine + Cisplatin + Y90 SIRT/SBRT Observational NCT05557578 Resection Tislelizumab + Gemcitabine + Oxaliplatin Phase II NCT05265208 Resection Capecitabine + Y90 SIRT Phase II NCT06910722 Transplantation Gemcitabine + Cisplatin+Y90 SIRT/SBRT Not applicable NCT06569225 Resection Gemcitabine + Cisplatin + nab-Paclitaxel + Rilvegostomig Phase II NCT06280508 Resection Adebrelimab + Apatinib + Gemcitabine + Cisplatin Phase II Abbreviations: HAIC, hepatic arterial infusion chemotherapy; SBRT, stereotactic body radiotherapy; SIRT, selective internal radiation therapy; TACE, transarterial chemoembolization.